Skip to Main Content
The realisation of a Walsh network is proposed. This network is built from a set of Walsh functions which are either odd or even binary value functions in their defined domain. The network is employed as an adaptive equaliser for which an algorithm for recursively adjusting the tap gain coefficients minimising the mean-square estimation error at the output is proposed. Some desirable properties of the covariance matrix of the equaliser are discussed. Examples of equalisation of a practical channel demonstrate the effectiveness of the Walsh equaliser.